Image analysis toolbox

The current image analysis code serves to analyse the chip microscopy images for the project presented in the pre-print titled A Multiscale Immuno-Oncology on-Chip System (MIOCS) establishes that collective T cell behaviors govern tumor regression. Each image is composed of a single spheroid-containing well and can be multi-channeled.

Installation

Make sure you use Python 3.8 by typing the command python -V.

Otherwise:

conda create -n ia python=3.8
conda activate ia

Install the package in development mode:

git clone git@gitlab.pasteur.fr:pub/image-analysis.git
cd image-analysis
pip install .

API

From the command line

The well images are extracted from their native ND2 format to tiff files using Nikon proprietary software. All the images are now in a single folder and indexed according to their position number, time frame and channel. See the format in the example_image folder.

The image analysis code is designed to be called from the command line in the terminal. To see the possible command options call:

python -m api.track_ot1 --help

Starting from the example images and using the default parameters, one types:

python -m api.track_ot1 PATH_TO_EXAMPLE_FOLDER

This should generate two new folders containing the segmentation results and the T-cell detection results, allowing the user to fine-tune the parameters for optimal detection results. This also generates a single CSV file that will be the basis for the future analysis.

The expected results from the segmentation are given in the folder expected_segmentation_output. It contains three items:

• A folder named Spheroid_Region_Detection containing another folder named 12 (for the well position index, a new folder is generated for each well), itself containg the segmentation results for the greyscale image showed to the left.
• A folder named verify_segmentation_OT1, also containing another folder named 12, itself containg the CTL detection results. The center of the spheroid is shown by the green dot, the CTL positions are given by the blue and red dots. The red color is used for cells detected as being on the spheroid whereas the blue dots are the CTLs detected in the gel.
• A CSV file named ot1_frame.csv recording all the CTL positions over time and relative to the spheroid.

From a Jupyter Notebook

After loading the virtual stack object, we analyze successively the spheroid and the T-cell track properties.

from api import read
from api import track

import matplotlib.pyplot as plt

save_path = r'/Example/Path/As/String'

vs = read.VirtualStack(save_path)

spheroid_frame = track.get_spheroid_properties(vs,
                                               spheroid_channel = 1, 
                                               fluo_channel = 3, 
                                               get_fluo = True, 
                                               verify_seg = True,
                                               wellSizeMu = 430,
                                               muTopx = 3)

t_cell_frame = track.get_cell_tracks(vs,
                                    fluo_channel = 2,
                                    muTopx = 3,
                                    search_range = 40,
                                    minsize = 10,
                                    minmass = 10000,
                                    percentile = 90)

You now have access to the spheroid and T-cell positions in your image stack.